<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
"http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head><meta http-equiv="content-type" content="text/html; charset=utf-8" />
<title>[192529] trunk/Source/JavaScriptCore</title>
</head>
<body>
<style type="text/css"><!--
#msg dl.meta { border: 1px #006 solid; background: #369; padding: 6px; color: #fff; }
#msg dl.meta dt { float: left; width: 6em; font-weight: bold; }
#msg dt:after { content:':';}
#msg dl, #msg dt, #msg ul, #msg li, #header, #footer, #logmsg { font-family: verdana,arial,helvetica,sans-serif; font-size: 10pt; }
#msg dl a { font-weight: bold}
#msg dl a:link { color:#fc3; }
#msg dl a:active { color:#ff0; }
#msg dl a:visited { color:#cc6; }
h3 { font-family: verdana,arial,helvetica,sans-serif; font-size: 10pt; font-weight: bold; }
#msg pre { overflow: auto; background: #ffc; border: 1px #fa0 solid; padding: 6px; }
#logmsg { background: #ffc; border: 1px #fa0 solid; padding: 1em 1em 0 1em; }
#logmsg p, #logmsg pre, #logmsg blockquote { margin: 0 0 1em 0; }
#logmsg p, #logmsg li, #logmsg dt, #logmsg dd { line-height: 14pt; }
#logmsg h1, #logmsg h2, #logmsg h3, #logmsg h4, #logmsg h5, #logmsg h6 { margin: .5em 0; }
#logmsg h1:first-child, #logmsg h2:first-child, #logmsg h3:first-child, #logmsg h4:first-child, #logmsg h5:first-child, #logmsg h6:first-child { margin-top: 0; }
#logmsg ul, #logmsg ol { padding: 0; list-style-position: inside; margin: 0 0 0 1em; }
#logmsg ul { text-indent: -1em; padding-left: 1em; }#logmsg ol { text-indent: -1.5em; padding-left: 1.5em; }
#logmsg > ul, #logmsg > ol { margin: 0 0 1em 0; }
#logmsg pre { background: #eee; padding: 1em; }
#logmsg blockquote { border: 1px solid #fa0; border-left-width: 10px; padding: 1em 1em 0 1em; background: white;}
#logmsg dl { margin: 0; }
#logmsg dt { font-weight: bold; }
#logmsg dd { margin: 0; padding: 0 0 0.5em 0; }
#logmsg dd:before { content:'\00bb';}
#logmsg table { border-spacing: 0px; border-collapse: collapse; border-top: 4px solid #fa0; border-bottom: 1px solid #fa0; background: #fff; }
#logmsg table th { text-align: left; font-weight: normal; padding: 0.2em 0.5em; border-top: 1px dotted #fa0; }
#logmsg table td { text-align: right; border-top: 1px dotted #fa0; padding: 0.2em 0.5em; }
#logmsg table thead th { text-align: center; border-bottom: 1px solid #fa0; }
#logmsg table th.Corner { text-align: left; }
#logmsg hr { border: none 0; border-top: 2px dashed #fa0; height: 1px; }
#header, #footer { color: #fff; background: #636; border: 1px #300 solid; padding: 6px; }
#patch { width: 100%; }
#patch h4 {font-family: verdana,arial,helvetica,sans-serif;font-size:10pt;padding:8px;background:#369;color:#fff;margin:0;}
#patch .propset h4, #patch .binary h4 {margin:0;}
#patch pre {padding:0;line-height:1.2em;margin:0;}
#patch .diff {width:100%;background:#eee;padding: 0 0 10px 0;overflow:auto;}
#patch .propset .diff, #patch .binary .diff {padding:10px 0;}
#patch span {display:block;padding:0 10px;}
#patch .modfile, #patch .addfile, #patch .delfile, #patch .propset, #patch .binary, #patch .copfile {border:1px solid #ccc;margin:10px 0;}
#patch ins {background:#dfd;text-decoration:none;display:block;padding:0 10px;}
#patch del {background:#fdd;text-decoration:none;display:block;padding:0 10px;}
#patch .lines, .info {color:#888;background:#fff;}
--></style>
<div id="msg">
<dl class="meta">
<dt>Revision</dt> <dd><a href="http://trac.webkit.org/projects/webkit/changeset/192529">192529</a></dd>
<dt>Author</dt> <dd>fpizlo@apple.com</dd>
<dt>Date</dt> <dd>2015-11-17 13:41:43 -0800 (Tue, 17 Nov 2015)</dd>
</dl>
<h3>Log Message</h3>
<pre>It's best for the DFG to always have some guess of basic block frequency
https://bugs.webkit.org/show_bug.cgi?id=151350
Reviewed by Geoffrey Garen.
It'll simplify things for B3 if we always have some estimate of block frequency, even if it's not
a great estimate. Using NaN as an estimate is probably worse than using any non-NaN number.
* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::inlineCall):
(JSC::DFG::ByteCodeParser::handleInlining):
(JSC::DFG::ByteCodeParser::parseCodeBlock):
* dfg/DFGCriticalEdgeBreakingPhase.cpp:
(JSC::DFG::CriticalEdgeBreakingPhase::breakCriticalEdge):
* dfg/DFGLoopPreHeaderCreationPhase.cpp:
(JSC::DFG::createPreHeader):</pre>
<h3>Modified Paths</h3>
<ul>
<li><a href="#trunkSourceJavaScriptCoreChangeLog">trunk/Source/JavaScriptCore/ChangeLog</a></li>
<li><a href="#trunkSourceJavaScriptCoredfgDFGByteCodeParsercpp">trunk/Source/JavaScriptCore/dfg/DFGByteCodeParser.cpp</a></li>
<li><a href="#trunkSourceJavaScriptCoredfgDFGCriticalEdgeBreakingPhasecpp">trunk/Source/JavaScriptCore/dfg/DFGCriticalEdgeBreakingPhase.cpp</a></li>
<li><a href="#trunkSourceJavaScriptCoredfgDFGLoopPreHeaderCreationPhasecpp">trunk/Source/JavaScriptCore/dfg/DFGLoopPreHeaderCreationPhase.cpp</a></li>
</ul>
</div>
<div id="patch">
<h3>Diff</h3>
<a id="trunkSourceJavaScriptCoreChangeLog"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/ChangeLog (192528 => 192529)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/ChangeLog 2015-11-17 21:31:57 UTC (rev 192528)
+++ trunk/Source/JavaScriptCore/ChangeLog 2015-11-17 21:41:43 UTC (rev 192529)
</span><span class="lines">@@ -1,3 +1,22 @@
</span><ins>+2015-11-17 Filip Pizlo <fpizlo@apple.com>
+
+ It's best for the DFG to always have some guess of basic block frequency
+ https://bugs.webkit.org/show_bug.cgi?id=151350
+
+ Reviewed by Geoffrey Garen.
+
+ It'll simplify things for B3 if we always have some estimate of block frequency, even if it's not
+ a great estimate. Using NaN as an estimate is probably worse than using any non-NaN number.
+
+ * dfg/DFGByteCodeParser.cpp:
+ (JSC::DFG::ByteCodeParser::inlineCall):
+ (JSC::DFG::ByteCodeParser::handleInlining):
+ (JSC::DFG::ByteCodeParser::parseCodeBlock):
+ * dfg/DFGCriticalEdgeBreakingPhase.cpp:
+ (JSC::DFG::CriticalEdgeBreakingPhase::breakCriticalEdge):
+ * dfg/DFGLoopPreHeaderCreationPhase.cpp:
+ (JSC::DFG::createPreHeader):
+
</ins><span class="cx"> 2015-11-17 Michael Saboff <msaboff@apple.com>
</span><span class="cx">
</span><span class="cx"> Reserved VM pool established in r187125 is likely too conservative
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoredfgDFGByteCodeParsercpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/dfg/DFGByteCodeParser.cpp (192528 => 192529)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/dfg/DFGByteCodeParser.cpp 2015-11-17 21:31:57 UTC (rev 192528)
+++ trunk/Source/JavaScriptCore/dfg/DFGByteCodeParser.cpp 2015-11-17 21:41:43 UTC (rev 192529)
</span><span class="lines">@@ -1505,7 +1505,7 @@
</span><span class="cx"> ASSERT(lastBlock->terminal());
</span><span class="cx">
</span><span class="cx"> // Need to create a new basic block for the continuation at the caller.
</span><del>- RefPtr<BasicBlock> block = adoptRef(new BasicBlock(nextOffset, m_numArguments, m_numLocals, PNaN));
</del><ins>+ RefPtr<BasicBlock> block = adoptRef(new BasicBlock(nextOffset, m_numArguments, m_numLocals, 1));
</ins><span class="cx">
</span><span class="cx"> // Link the early returns to the basic block we're about to create.
</span><span class="cx"> for (size_t i = 0; i < inlineStackEntry.m_unlinkedBlocks.size(); ++i) {
</span><span class="lines">@@ -1848,7 +1848,7 @@
</span><span class="cx">
</span><span class="cx"> for (unsigned i = 0; i < callLinkStatus.size(); ++i) {
</span><span class="cx"> m_currentIndex = oldOffset;
</span><del>- RefPtr<BasicBlock> block = adoptRef(new BasicBlock(UINT_MAX, m_numArguments, m_numLocals, PNaN));
</del><ins>+ RefPtr<BasicBlock> block = adoptRef(new BasicBlock(UINT_MAX, m_numArguments, m_numLocals, 1));
</ins><span class="cx"> m_currentBlock = block.get();
</span><span class="cx"> m_graph.appendBlock(block);
</span><span class="cx"> prepareToParseBlock();
</span><span class="lines">@@ -1899,7 +1899,7 @@
</span><span class="cx"> }
</span><span class="cx">
</span><span class="cx"> RefPtr<BasicBlock> slowPathBlock = adoptRef(
</span><del>- new BasicBlock(UINT_MAX, m_numArguments, m_numLocals, PNaN));
</del><ins>+ new BasicBlock(UINT_MAX, m_numArguments, m_numLocals, 1));
</ins><span class="cx"> m_currentIndex = oldOffset;
</span><span class="cx"> m_exitOK = true;
</span><span class="cx"> data.fallThrough = BranchTarget(slowPathBlock.get());
</span><span class="lines">@@ -1933,7 +1933,7 @@
</span><span class="cx"> }
</span><span class="cx">
</span><span class="cx"> RefPtr<BasicBlock> continuationBlock = adoptRef(
</span><del>- new BasicBlock(UINT_MAX, m_numArguments, m_numLocals, PNaN));
</del><ins>+ new BasicBlock(UINT_MAX, m_numArguments, m_numLocals, 1));
</ins><span class="cx"> m_graph.appendBlock(continuationBlock);
</span><span class="cx"> if (verbose)
</span><span class="cx"> dataLog("Adding unlinked block ", RawPointer(continuationBlock.get()), " (continuation)\n");
</span><span class="lines">@@ -4895,7 +4895,7 @@
</span><span class="cx"> m_currentBlock = m_graph.lastBlock();
</span><span class="cx"> m_currentBlock->bytecodeBegin = m_currentIndex;
</span><span class="cx"> } else {
</span><del>- RefPtr<BasicBlock> block = adoptRef(new BasicBlock(m_currentIndex, m_numArguments, m_numLocals, PNaN));
</del><ins>+ RefPtr<BasicBlock> block = adoptRef(new BasicBlock(m_currentIndex, m_numArguments, m_numLocals, 1));
</ins><span class="cx"> m_currentBlock = block.get();
</span><span class="cx"> // This assertion checks two things:
</span><span class="cx"> // 1) If the bytecodeBegin is greater than currentIndex, then something has gone
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoredfgDFGCriticalEdgeBreakingPhasecpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/dfg/DFGCriticalEdgeBreakingPhase.cpp (192528 => 192529)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/dfg/DFGCriticalEdgeBreakingPhase.cpp 2015-11-17 21:31:57 UTC (rev 192528)
+++ trunk/Source/JavaScriptCore/dfg/DFGCriticalEdgeBreakingPhase.cpp 2015-11-17 21:41:43 UTC (rev 192529)
</span><span class="lines">@@ -73,9 +73,7 @@
</span><span class="cx"> private:
</span><span class="cx"> void breakCriticalEdge(BasicBlock* predecessor, BasicBlock** successor)
</span><span class="cx"> {
</span><del>- // Note that we pass NaN for the count of the critical edge block, because we honestly
- // don't know its execution frequency.
- BasicBlock* pad = m_insertionSet.insertBefore(*successor, PNaN);
</del><ins>+ BasicBlock* pad = m_insertionSet.insertBefore(*successor, (*successor)->executionCount);
</ins><span class="cx"> pad->appendNode(
</span><span class="cx"> m_graph, SpecNone, Jump, (*successor)->at(0)->origin, OpInfo(*successor));
</span><span class="cx"> pad->predecessors.append(predecessor);
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoredfgDFGLoopPreHeaderCreationPhasecpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/dfg/DFGLoopPreHeaderCreationPhase.cpp (192528 => 192529)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/dfg/DFGLoopPreHeaderCreationPhase.cpp 2015-11-17 21:31:57 UTC (rev 192528)
+++ trunk/Source/JavaScriptCore/dfg/DFGLoopPreHeaderCreationPhase.cpp 2015-11-17 21:41:43 UTC (rev 192529)
</span><span class="lines">@@ -64,10 +64,17 @@
</span><span class="cx"> // SSACalculator and treating the Upsilons as Defs and rebuilding the Phis from scratch.
</span><span class="cx"> //
</span><span class="cx"> // https://bugs.webkit.org/show_bug.cgi?id=148587
</span><del>-
- // Don't bother to preserve execution frequencies for now.
- BasicBlock* preHeader = insertionSet.insertBefore(block, PNaN);
</del><span class="cx">
</span><ins>+ // Determine a good frequency for the pre-header. It's definitely not the frequency of the loop body.
+ // Instead, we use the max of the frequencies of the loop body's non-loop predecessors.
+ float frequency = 0;
+ for (BasicBlock* predecessor : block->predecessors) {
+ if (graph.m_dominators->dominates(block, predecessor))
+ continue;
+ frequency = std::max(frequency, predecessor->executionCount);
+ }
+ BasicBlock* preHeader = insertionSet.insertBefore(block, frequency);
+
</ins><span class="cx"> // FIXME: It would be great if we put some effort into enabling exitOK at this origin, if it
</span><span class="cx"> // happens to be unset. It might not be set because the loop header (i.e. "block") has Phis in it.
</span><span class="cx"> // Phis have to have exitOK=false. There are a few ways to try to set exitOK:
</span></span></pre>
</div>
</div>
</body>
</html>